Change status code from 500 to 400 for well-defined errors (if possible to do cleanly)

[stroxler]

It looks like two different well-defined user errors result in a status code of 500

• specifying a nonexistent docker image
• trying to run on a nonexistent ECS cluster

Ideally, we would report a 400 on these user errors. A caveat is that I'm not sure that's possible to do cleanly unless flotilla keeps track of the permitted images and clusters, since failing to find an existing docker image or flotilla cluster really should result in a 500 rather than a 400